图像雅可比矩阵伪逆估计在视觉伺服中的应用  

Application of Pseudo-inverse Estimation of Image Jacobian Matrix to Visual Servoing

作　　者：康庆生[1] 孟正大[1] 戴先中[1] 

机构地区：[1]东南大学自动控制系

出　　处：《机器人》2006年第4期406-409,414,共5页Robot

基　　金：国家863计划资助项目(2003AA420010);江苏省十五科技攻关计划资助项目(BE200203114)

摘　　要：在一般无标定视觉伺服算法的基础上,提出了一种基于递推最小二乘法估计图像雅可比矩阵伪逆的无标定视觉伺服算法.这种方法无需计算雅可比矩阵伪逆.针对固定目标,文中给出了比例控制算法的稳定性条件.针对运动目标,推导了基于图像雅可比矩阵伪逆估计的伪高斯—牛顿视觉伺服算法和信任区视觉伺服算法.最后,仿真验证了三种视觉伺服算法的有效性.Based on the general uncalibrated visual servoing algorithm, this paper presents a new visual servoing control scheme which estimates pseudo-inverse of image Jacobian matrix based on recursive least square algorithm. This method doesn＇t compute the Moore-Penrose inverse of image Jacobian. Stability condition of the proportional control algorithm is given for a fixed target. For moving targets, the quasi-Gauss-Newton and trust region visual servoing control strategies are deduced based on pseudo-inverse estimation of image Jacobian matrix. In the end, the performance of three visual servoing algorithms is validated by simulations.

关 键 词：伪逆 图像雅可比矩阵 估计 视觉伺服 无标定 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]